Overestimating the Salience of Racial Labels

Abstract

In the present study, a word at the end of one page was wrongly doubled at the top of the next page. The word was white, black, or some other label applied to men. Racial labels were no more likely to elicit the detection of doubling errors than other labels, according to a baseline experiment on 352 subjects divided into five treatment groups. In Experiment 2, 134 South Africans, most of whom were white, estimated what the percentage of error detectors had been for each doubled label. Detection of errors in racial labels was more likely to be overestimated than detection of errors in other labels. An assumption by the subjects that race is as salient in the cognitions of others as in their own cognitions may explain the over-estimation.